Fax: (713) 745-4286
Feasibility of an expressive-disclosure group intervention for post-treatment colorectal cancer patients
Results of the Healthy Expressions study
Article first published online: 8 APR 2011
Copyright © 2011 American Cancer Society
Volume 117, Issue 21, pages 4993–5002, 1 November 2011
How to Cite
Carmack, C. L., Basen-Engquist, K., Yuan, Y., Greisinger, A., Rodriguez-Bigas, M., Wolff, R. A., Barker, T., Baum, G. and Pennebaker, J. W. (2011), Feasibility of an expressive-disclosure group intervention for post-treatment colorectal cancer patients. Cancer, 117: 4993–5002. doi: 10.1002/cncr.26110
- Issue published online: 19 OCT 2011
- Article first published online: 8 APR 2011
- Manuscript Accepted: 24 JAN 2011
- Manuscript Revised: 14 DEC 2010
- Manuscript Received: 16 JUL 2010
- colorectal cancer;
- psychosocial intervention;
- psychological distress;
- group support;
- expressive disclosure
Adjusting to cancer requires effective cognitive and emotional processing. Written and verbal disclosure facilitate processing and have been studied independently in cancer survivors. Combined written and verbal expression may be more effective than either alone, particularly for patients with difficult to discuss or embarrassing side effects. Thus, the authors developed and tested the efficacy of a 12-session combined written and verbal expression group program for psychologically distressed colorectal cancer (CRC) patients.
Forty post-treatment patients with CRC (stages I-III) identified as psychologically distressed using the Brief Symptom Inventory (BSI) were randomized to an intervention group (Healthy Expressions; n = 25) or standard care (control group; n = 15). Assessments were completed at baseline, Month 2, and Month 4 (postintervention). Primary outcomes were psychological functioning and quality of life (QOL).
Most participants were women (63%), white (63%), and non-Hispanic (75%). The Healthy Expressions group demonstrated significantly greater changes in distress compared with the control group at Month 2 on the BSI Global Severity Index (GSI) and the Centers for Epidemiologic Studies Depression scale (CES-D) scores (P < .05 for each); differences in the European Organization for Research and Treatment of Cancer (EORTC) global QOL scores approached significance (P = .063). The BSI GSI and Positive Symptom Total, CES-D, and EORTC emotional functioning subscale scores were all significant at Month 4 (P < .05 for each).
The Healthy Expressions program improved psychological functioning in CRC patients who reported experiencing distress. Findings demonstrate the program's feasibility and provide strong support for conducting a larger randomized trial. Cancer 2011;. © 2011 American Cancer Society.
Patients with colorectal cancer (CRC) often have quality of life (QOL) concerns, such as those relating to stomas and post-treatment sexual function.1-3 Concerns also arise from unanticipated or severe physical symptoms, including fatigue, pain, difficulty sleeping, shortness of breath, and diarrhea.4-6 These enduring physical changes can cause psychological distress,7-10 including depression and anxiety.8 Such physical and psychological problems can disrupt routine social activities. Symptoms including pain, fatigue, anxiety, and embarrassment can impact interactions with family and friends.9, 11 Patients may believe that others do not know what to say, so they avoid speaking about their distress12, 13; however, avoidance and limited disclosure may have a negative effect.14, 15
According to the social cognitive processing model, talking with others facilitates one's processing of cancer if the social context for discussions is supportive and positive. This cognitive processing can be inhibited in the presence of social constraints: limited or no access to supportive others or having a support network that does not know how to respond when cancer is discussed.14 Because CRC patients experience side effects that may be difficult or embarrassing to discuss, we developed a program to promote cognitive processing, an expressive-disclosure support group program.
An expressive disclosure group program is a low-cost, easy-to-deliver intervention for cancer patients and may include written and/or verbal expression. Disclosing or expressing thoughts and feelings about stressful events improves physical and mental health.16-18 These interventions may be effective because they direct patients' attention to sources of stress, promote habituation to these experiences through repeated exposure, and facilitate cognitive restructuring of stressful situations.19
Written disclosure is associated with improved health, psychological well-being, physiological functioning, and general functioning.16-18, 20-22 Whether expressive writing can reduce psychological distress in cancer patients requires further study,23 as most of this research with cancer patients has not found changes in distress.23-26 In fact, expressive writing initially causes short-term increases in distress,15, 18, 27 which benefits some. For others, this initial distress could increase avoidance. Patients facing cancer, especially those who use avoidant coping, may benefit more from intensive and prolonged emotional disclosure interventions than from the typically brief expressive writing interventions.28 Verbal disclosure with written expression, in the presence of a supportive group leader17 and fellow group members, could reduce distress, prevent avoidant coping, and reduce intrusive thoughts. In fact, verbal disclosure reduces intrusive thoughts,29, 30 particularly when disclosure is validated.30 Thus, the combination of expressive writing and talking may be more therapeutic than either alone.17
In a combined approach, brief expressive writing before discussion can initiate emotional and cognitive processing as thoughts are organized into a coherent narrative, and can facilitate group sharing by priming patients before discussion.31 Cognitive processing initiated during writing can continue through discussion. For those who use avoidant coping, the group process encourages continued disclosure, thus reinforcing active coping. Talking with and listening to the group also allows patients to gain outside opinions, receive support, and obtain coping information.15 As group members suggest alternative ways of coping or positively reframe a situation, they facilitate cognitive restructuring and processing stressful events. Expressive writing after the discussion provides opportunity for patients to reflect and cognitively integrate the ideas discussed.
We present results from a small randomized pilot trial to test the feasibility of this program (called Healthy Expressions) and compare its efficacy to standard care in 40 patients with CRC who were experiencing psychological distress. We hypothesized that patients randomized to Healthy Expressions would show greater changes in psychological functioning and QOL than those in standard care, as well as greater changes in cognitive processing, coping strategies, and social support.
MATERIALS AND METHODS
Forty participants from 3 cohorts of study recruits were randomly assigned to 2 groups: the Healthy Expressions group (intervention group) or standard care (control group).
Eligible patients had a diagnosis of stage I, II, or III colon or rectal cancer; had completed treatment; were distressed (Brief Symptom Inventory [BSI]32, 33 clinical cutoff for distress with a T score ≥63 on the Global Severity Index [GSI] or a T score ≥63 on any 2 of the 9 primary symptom dimensions); could read, speak, and write English; resided within 1 hour of The University of Texas MD Anderson Cancer Center; were 18 years old or older; and could provide informed consent.
Stage IV patients were excluded, as their QOL concerns include end of life issues not faced by those with earlier stage disease. We excluded patients in active treatment, as attending group sessions during treatment may burden patients and reduce the likelihood of sustained participation. There were no requirements with regard to time since diagnosis or end of treatment.
This research was conducted after approval from The University of Texas MD Anderson Cancer Center Institutional Review Board. Patients were recruited from The University of Texas MD Anderson Cancer Center and Kelsey-Seybold Clinic. At The University of Texas MD Anderson Cancer Center, potentially eligible participants, identified from clinic appointment schedules, either received an invitation letter before their next appointment or were directly approached during their next appointment. Those who received the letter were provided a telephone number to call if they were not interested.
At Kelsey-Seybold Clinic, potentially eligible patients, identified from a patient database, were sent informational letters about the study. Patients interested in participating returned a postage-paid card providing permission for The University of Texas MD Anderson Cancer Center staff to contact them. Kelsey-Seybold Clinic provided 1 follow-up phone call to those not responding. The University of Texas MD Anderson Cancer Center staff contacted patients who returned the card to conduct an initial eligibility screening.
At both study sites, The University of Texas MD Anderson Cancer Center staff met with patients either during their next medical appointment or at a time of their choosing. Patients who provided informed consent were screened with the BSI, and those identified as distressed were asked to participate. We recruited participants in cohorts, or groups of approximately 15 patients. These cohorts completed the baseline assessment and then were randomized to study condition.
Study Group Assignment
Participants were assigned to the study groups using minimization,34, 35 which is similar to stratified randomization but results in better group balance and does not have some of stratification's limitations (eg, increased probability of group imbalance when stratifying across a large number of participant characteristics). In minimization, before treatment condition assignment, the number of participants already in each treatment condition with similar characteristics is totaled. These totals are computed based on each participant characteristic considered separately. A patient is then assigned to a treatment condition based on what would yield the best overall balance. Published simulation studies36 show that the procedure works well with small samples. For this study, age, ethnicity, sex, time since diagnosis, disease stage, disease site (colon vs rectal), number of medical comorbidities, baseline BSI score, and use of psychotropic medication were used during minimization.
Standard care control group participants did not attend formal support groups other than those attended as a part of standard care. They received 2 mailings of educational and community resource information related to support needs of cancer patients.
Healthy Expressions intervention group participants attended group sessions with approximately 8 members. Throughout the program, they received the same educational and informational material as the control group. The intervention included 12 one-hour sessions over 4 months (9 weekly sessions, 2 bimonthly sessions, and 1 concluding session in Month 4) facilitated by 2 Master's level interventionists. The intervention involved disclosure through written expression and facilitated discussion covering topics identified through a prior formative evaluation with 20 CRC patients who did not participate in this trial; they were individually interviewed to assess QOL issues they would like to discuss in a group format and to ascertain their interest in a CRC support group. Identified topics which were covered in the group included fatigue management, interacting with the health care team, talking with family and friends about cancer, persistent physical symptoms, bowel management, sexuality, spirituality, emotional well-being, body image, and fear of recurrence. A structured curriculum with a detailed manual for evaluating the fidelity of treatment delivery across cohorts was used. Participants provided informed consent for all sessions to be audiorecorded. The first author reviewed approximately 20% of randomly selected tapes for quality control, including verifying the time spent writing and discussing, and ensuring session topics were covered. Copies of the tapes were provided to participants who missed a session.
For the first 5 minutes of each session, participants wrote about the session topic in a journal. The writing instructions were phrased to elicit cognitions and emotions, both of which are needed to receive benefit from expressive writing.37 Participants were asked to put away their writings before the discussion, keeping the journals either in their possession or in a study lock box. They were told that they did not have to disclose any of their writings unless they so chose, because privacy during expressive writing results in better psychological outcomes.20
Next, for 30 minutes, participants engaged in a facilitated discussion, using pre-established questions to elicit the discussion; pre-established questions ensured continuity of intervention delivery across groups. The interventionists posed essential questions for each session; if time remained, additional questions were available. For example, for the session on sexuality, 2 essential questions were “What are your thoughts and feelings about your sexuality, including any changes in your desire or ability to have sex?” and “How have you coped with the changes in your desire or ability to have sex?” All questions were worded to promote cognitive and emotional processing and complemented the writing instructions.
After the discussion, participants wrote in their journals for 20 minutes using the same instructions as the first writing exercise. Participants were also asked to consider some of the issues raised during the discussion.
Participants completed questionnaires administered in person at baseline and at Month 4 (postintervention) and by mail at Month 2 (interim assessment). Participants received a $10 gift card for each completed assessment. Research staff who administered the questionnaires were blinded to the study group assignments.
At baseline, participants provided their age, sex, race/ethnicity, marital status, occupational status, educational history, diagnosis date, disease stage, disease site, treatment, medical comorbidities, and current medications.
Primary outcome measures
Psychological functioning was assessed using the BSI32 and the Centers for Epidemiologic Studies Depression scale (CES-D).38 A CES-D score ≥16 indicates the need for further clinical evaluation. QOL was assessed using the European Organization for Research and Treatment of Cancer (EORTC) QOL questionnaire,39 which comprises 5 functional scales (including emotional functioning), 3 symptom scales, and a global health and QOL scale. Because of our small sample, we limited our analysis to the EORTC global scale and the emotional functioning scale.
Mediating variable measures
Cognitive processing was measured using the Impact of Events Scale (IES)40, 41 to assess intrusive and avoidant thoughts related to CRC and the revised Automatic Thoughts Questionnaire42, 43 to assess the frequency of negative and positive self-statements associated with depression and other psychopathology. Coping was measured using a 24-item version of the COPE, which instructed participants to consider their responses to CRC.44 Although the COPE has 12 distinct subscales, to minimize the number of analyses, only the subscales related to support (ie, use of emotional support) and cognitive processing (ie, positive reframing and planning) were examined. Social support was measured using the Interpersonal Support Evaluation List (ISEL).45, 46 A total score and 4 subscale scores were derived; only the total score was examined. In addition, 9 items from the Cancer Rehabilitation Evaluation System47 were used to assess perceived social constraints in talking about cancer with friends and family and spouses.48
Process evaluation measures
To measure the recruitment rate, we tracked the number of participants contacted, the number eligible, and the number who consented. Participants who were eligible (including those who declined) provided demographic and medical information, and nonparticipants provided their reasons for declining. Retention was calculated as the percentage completing each assessment session; reasons for study withdrawal were also recorded. To assess contamination, at Month 4, participants were asked about other activities they participated in outside of this study to enhance QOL (eg, support groups, counseling). To evaluate intervention exposure, we tracked session attendance. Intervention group participants completed participant satisfaction evaluations at the end of each session and at the conclusion of the program.
The primary outcomes were the Month 4 psychological functioning and QOL scores. Forty participants randomized into the control and intervention arms provided 80% power (1-sided alpha = .05) to detect an effect size of 0.81 standard deviation (SD) difference between the 2 arms, assuming 10% attrition.
We calculated descriptive statistics and used an intent-to-treat approach to compare outcomes between the 2 conditions. Because this study used a repeated measures design, a linear mixed model was used for data analysis. In the model, the interventions, measurements, and their interactions were included as covariates, and random intercepts were used to account for within-subject correlations. At each time point, comparisons between the 2 arms on the outcome and mediator variables were conducted based on contrasts using SAS Proc Mixed. This approach is more powerful than conducting separate tests at each time point (eg, based on the t test), because it uses all data information for inference. We estimated the mediation effect based on the framework of Baron and Kenny.49 Given the small sample and nature of the study, formal tests of mediation were not conducted.
Recruitment and Study Sample
CRC patients were recruited from 2 sites in Houston: The University of Texas MD Anderson Cancer Center and Kelsey-Seybold Clinic. At The University of Texas MD Anderson Cancer Center, 289 patients were directly approached during their clinic appointment. Of these, 59% (n = 169) declined and may or may not have been eligible, and 5% (n = 15) were ineligible. Most declined because of lack of time (n = 97), not wanting to come to The University of Texas MD Anderson Cancer Center for group therapy (n = 27), lack of interest (n = 14), and transportation issues (n = 12). A total of 36% (n = 105) consented to be screened for distress for eligibility; 52 were eligible (49.5% of those consenting). This rate of distress is higher than published rates (ie, 31.6%8) and likely reflects the finding that nondistressed patients declined screening, knowing they would not be eligible and the group would not benefit them. At Kelsey-Seybold Clinic, Health Insurance Portability and Accountability Act guidelines required a passive recruitment approach. After Kelsey-Seybold Clinic sent letters to 375 patients and followed up once by phone, 38 patients returned consent forms to release their information to our staff. Of these, 45% (n = 17) declined, and 16% (n = 6) were ineligible. Many declined because of lack of time (n = 6). A total of 39% (n = 15) consented to be screened for distress; 11 were eligible (73.3% of those consenting). See Figure 1 for recruitment details at The University of Texas MD Anderson Cancer Center and Kelsey-Seybold Clinic.
From both sites, 120 consented to be screened, and 63 were eligible. Of these, 23 withdrew after BSI screening and before completing the baseline assessment. Reasons for withdrawal included work (n = 17), gas/driving (n = 4), and illness (n = 2). We compared participants and nonparticipants (both those refusing and those not eligible) to determine whether selection bias was present. Those who agreed to participate were much younger (mean age, 57.90 years; SD, 10.70) than those who refused (mean age, 63.64 years; SD, 11.00); among those who agreed to be screened, those who were eligible were younger (mean age, 56.09 years; SD, 10.04) than those who were not eligible (mean age, 60.13 years; SD, 10.20). There were no significant differences in sex, ethnicity, time since diagnosis, year treatment was completed, cancer diagnosis (colon vs rectal), or disease stage.
Most of the 40 participants who completed the baseline assessments were married (60%), women (63%), employed full or part time (51%), and college educated (51%). Participants were 63% white, 18% African American, 15% Hispanic, and 5% Asian. Mean age was 56.16 years (SD, 10.22), with a range of 27 to 75 years. Mean time since diagnosis was 45.30 months (SD, 39.02), with a range of 11 to 214 months. Disease stage at diagnosis was 22% stage I, 18% stage II, and 60% stage III. All participants had undergone surgery, 53% had received radiation therapy, and 74% had received chemotherapy. There were no significant differences in the demographic and medical variables between the intervention and control groups.
Barriers to randomization
Our original intent was for each study condition to have equal numbers of participants per cohort. Because we wanted to ensure that the group program would have at least 8 members, <8 were assigned to standard care per cohort. Ultimately, 25 were assigned to the intervention group and 15 to the control group.
Other Process Evaluation Data
Attendance and barriers to participation
In the intervention group, mean attendance was 7.12 sessions (of 12); 17 (of 25) participants attended at least 6 sessions, and 3 participants attended none. Sessions for all cohorts were held over lunch, which was provided. The main reasons for absences included work (42%), illness/injury (18%), and travel (8.2%). Also, Hurricane Ike hit before session 11 of the third cohort and adversely affected participation for that session.
There were no differences between the study groups in seeking other programs outside the study to improve QOL; 31% in the intervention group (n = 5) and 30% in the control group (n = 3) sought other services (P = .648).
The mean ratings on the Healthy Expressions program evaluation indicated that the content was “more than expected.” On a 5-rating scale with 5 being the highest rating, ratings were high with regard to positive feelings toward group members (mean, 3.89; SD, 0.74), comfort in talking openly in the group (mean, 4.42; SD, 0.96), and willingness to discuss CRC cancer with others (mean, 4.74; 5 = more willing; SD, 0.56). Participants reported they would definitely recommend the program to other patients (mean, 4.89 of 5; SD, 0.46).
Participants rated session length and program duration (1 = too short, 3 = just right, and 5 = too long); ratings averaged 2.47 (SD, 0.61) and 2.79 (SD, 0.63) for session length and program duration, respectively. By using the same rating scale, participants rated the discussion as too short (mean, 2.32; SD, 0.67) and the writing as too long (mean, 3.95; SD, 0.91). Similarly, although ratings (1 = not at all useful; 3 = useful; 5 = extremely useful) indicated that participants preferred discussion (mean, 4.75; SD, 0.46) over writing (mean, 2.95; SD, 0.91), both were considered useful. Interestingly, 30% said they did some writing at home outside of the program.
At Month 2, we had complete data for 88% in the intervention group and 87% in the control group. At Month 4, we had complete data for 96% and 100% in the intervention and control groups, respectively.
Efficacy of Healthy Expressions
Table 1 compares effect sizes (mean/SD) between the 2 groups at Month 2 and 4 (adjusted for baseline) for the primary outcome and mediator variables. Table 2 provides estimates of the mediation effects at Month 4, with a focus only on the mediator and outcome variables that changed significantly from baseline to Month 4.
|Variable||Month 2||Month 4|
|Effect Size||Pa||Effect Size||Pa|
|EORTC emotional functioning||0.05||.767||1.96||.028b|
|Social constraints (friends and family)||1.11||.286||2.21||.022b|
|Social constraints (spouse)||1.03||.328||2.39||.048b|
|COPE use of emotional support||0.26||.550||0.65||.284|
|COPE positive reframing||0.24||.787||0.39||.570|
|BSI GSI||BSI PST||CES-D||EORTC Emotional Functioning|
|IES intrusion||2.31 (1.90)||2.50 (2.07)||3.07 (2.45)||−6.40 (5.10)|
|Social constraints (friends and family)||−1.55 (1.32)||−1.40 (1.32)||−2.41 (1.83)||1.96 (2.27)|
|Social constraints (spouse)||−1.95 (1.74)||−2.23 (1.95)||−2.50 (2.00)||1.03 (3.34)|
Primary outcome variables
At Month 2, the intervention group scored significantly better on BSI-GSI and CES-D; EORTC-Global QOL approached significance. At Month 4, the intervention group scored significantly better on the BSI-GSI (Fig. 2), BSI-PST, CES-D (Fig. 3), and EORTC emotional functioning subscale (Fig. 4). There were no significant differences on the BSI-PSDI.
Intervention group participants demonstrated significantly lower IES avoidance at Month 2 and IES intrusion at Month 4. Interestingly, they showed higher social constraints (friends and family) and social constraints (spouse) at Month 4. There were no significant differences on positive or negative thoughts (Automatic Thoughts Questionnaire), coping (COPE), or social support (Interpersonal Support Evaluation List).
Our intervention improved psychological functioning in distressed CRC patients compared with a standard care control condition. Our findings were consistent across multiple primary outcome measures and are impressive given our sample size. Our results were also clinically meaningful, as reflected by the mean CES-D score of <16 in the intervention group at Months 2 and 4, whereas means remained >16 for the control group.
Because the side effects of CRC may negatively affect patients' social interactions, thus impacting their social support, group programs focusing on expressive disclosure may be an effective forum for improving their psychological functioning. Disclosure often results in positive social interactions, where the patient is reinforced with validation and acceptance. Listening to and talking with others may also promote cognitive reappraisal and help the patient to find meaning in his/her illness.50 We proposed that cognitive processing may mediate intervention effects on psychological distress. Changes in 1 measure of cognitive processing—intrusive and avoidant thoughts—indicated this may be a possibility. A future follow-up study that is adequately powered to test mediation will be necessary to properly answer this question.
We proposed that social support also may mediate intervention efficacy. Interestingly, intervention group participants showed increased social constraints in talking about cancer to their friends, family, and spouses. These results are similar to Helgeson and colleagues' findings showing that participants with more social support demonstrated worse physical functioning QOL after participating in a peer support program.51 In both studies, participants may have recognized how supportive the peer group is in comparison to their own social network, thus causing a negative reappraisal of their own network. A future follow-up study is required to further examine the role of social support in intervention efficacy.
Finally, we found no significant changes in patients' coping strategies, specifically coping strategies related to emotional support and cognitive processing. A future trial to study a wider range of coping strategies may yield different results. Conversely, coping may be dispositional and therefore resistant to change. Although a recent meta-analysis indicated that expressive writing alone does not improve coping,20 there is evidence that it may be most beneficial for those with negative coping styles.52
This study has several strengths, setting the stage for a subsequent larger study. First, our intervention was based on a strong theoretical foundation that allows for the examination of the mechanisms through which changes occur in the outcome variables.53 Second, we developed a support program for CRC patients, an understudied population with respect to psychosocial interventions. Third, we limited the study to patients experiencing psychological distress; most studies include all patients without regard to their need for intervention, which is not cost-effective. Fourth, our intervention does not require advanced training for its dissemination and can be administered by health educators who conduct traditional support groups. Finally, our detailed program manual will promote the evaluation of treatment fidelity, allow comparison across studies, and enable future widespread dissemination.
The study is limited by the lack of an attention control group. Given the minimal nature of our intervention, at most, such a group might include the discussion only component of our intervention. This would provide information as to whether the expressive writing component adds to a discussion only format, the format typically provided in traditional support groups. A future 3-arm trial could determine whether the combined approach is more effective than either alone. A future trial may also consider delivery of the intervention through a distance-based approach, which may assist with overcoming some of the barriers to recruitment and session attendance (eg, time, travel) and with future widespread dissemination. The study is also limited by its small sample size and lack of follow-up data. However, results are promising and provide strong support for conducting a larger randomized trial to further examine intervention efficacy, maintenance of treatment effects, and potential moderators of treatment efficacy, such as time since diagnosis. Despite these limitations, our pilot study is an important first step in demonstrating the feasibility and efficacy of an easy-to-deliver psychosocial intervention for distressed CRC patients.
This research was funded by a grant from the National Cancer Institute (CA107306) to Cindy L. Carmack, PhD. Support was also provided by the Patient-Reported Outcomes, Survey, and Population Research (PROSPR) shared resource (National Cancer Institute CA16672).
CONFLICT OF INTEREST DISCLOSURES
The authors made no disclosures.
- 13Face, honor and dignity in the context of colon cancer. J Med Humanit. 2000; 21: 229-243., , , , .
- 16Writing about emotional experiences as a therapeutic process. Psychol Sci. 1997; 8: 162-166..Direct Link:
- 19Expressive writing and health: self-regulation of emotion-related experience, physiology, and behavior. In: Lepore SJ, Smyth JM, eds. The Writing Cure: How Expressive Writing Promotes Health and Emotional Well-Being. Washington, DC: American Psychological Association; 2002: 99-117., , , .
- 23Emotional expression, expressive writing, and cancer. In: Lepore SJ, Smyth JM, eds. The Writing Cure: How Expressive Writing Promotes Health and Emotional Well-Being. Washington, DC: American Psychological Association; 2002: 31-51., .
- 26Expressive writing as a presurgical stress management intervention for breast cancer patients. J Soc Integr Oncol. 2008; 6: 59-66., , , et al.
- 31Group processes in therapeutic support groups. In: Baum A, Andersen BL, eds. Psychosocial Interventions for Cancer. Washington, DC: American Psychological Association; 2001: 131-141..
- 32Brief Symptom Inventory: Administration, Scoring and Procedures Manual. 3rd ed. Minneapolis, MN: National Computer Systems; 1993..
- 34Fundamentals of Clinical Trials. ed. 3rd ed. New York, NY: Springer-Verlag; 1998., , .
- 35Clinical Trials: A Practical Approach. New York, NY: John Wiley & Sons; 1983..
- 38The CES-D scale: a self-report depression scale for research in the general population. Appl Psychol Meas. 1977; 1: 385-401..
- 45Measuring the functional components of social support. In: Sarason IG, Sarason BR, eds. Social Support: Theory, Research, and Application. The Hague, Holland: Martinus Nijhoff; 1985: 73-94., , , .
- 50A social-cognitive processing model of emotional adjustment to cancer. In: Baum A, Andersen BL, eds. Psychosocial Interventions for Cancer. Washington, DC: American Psychological Association; 2001: 99-116..